Ideally we would like experimental subjects to be perfect strangers so that the situation they face at the lab is not just part of a long run interaction. Unfortunately, it is not easy to reach those conditions and experimenters try to mitigate any e?ects from these out-of-the-lab relationships by, for instance, randomly matching subjects. However, even if this type of procedure is used, it cannot be excluded that a subject may face a friend or an acquaintance. For the dictator game we find evidence that a positive probability of playing with a friend is not relevant to experimental results. However, when subjects are certain to face a friend the results change.